Cluster_2D_Visualization.m

% This script uses the k-means function on an example with randomly created
% data points in 2D and visualaizes its result.

close all; clear; clc;


%% Input args (and print them)

k = 10;     % number of clusters
numP = 200; % number of points
xMax = 100; % x between 0 and xMax
yMax = 100; % y between 0 and yMax
fprintf('k-Means will run with %d clusters and %d data points.\n',k,numP);


%% Create random data points

% create a random matrix of size 2-by-numP 
% with row 1/2 (x/y coordiante) ranging in [minX,maxX]/[minY,maxY]
xP = xMax * rand(1,numP);
yP = yMax * rand(1,numP);
points = [xP; yP];


%% run kMeans.m and measure/print performance

tic;
[cluster, centr] = kMeans(k, points); % my k-means
myPerform = toc;
fprintf('Computation time for kMeans.m: %d seconds.\n', myPerform);


%% run MATLAB's function kmeans(P,k) and measure/print performance

tic;
[cluster_mT, centr_m] = kmeans(points',k); % MATLAB's k-means
matlabsPerform = toc;
cluster_m = cluster_mT';
fprintf('Computation time for MATLABs kmeans: %d seconds.\n', matlabsPerform);


%% Compare performances

frac = matlabsPerform/myPerform;
fprintf('MATLAB uses %d of the time kMeans.m uses.\n' ,frac);


%% All visualizations

figure('Name','Visualizations','units','normalized','outerposition',[0 0 1 1]);

% visualize the clustering
subplot(2,2,1);
scatter(xP,yP,200,cluster,'.');
hold on;
scatter(centr(1,:),centr(2,:),'xk','LineWidth',1.5);
axis([0 xMax 0 yMax]);
daspect([1 1 1]);
xlabel('x');
ylabel('y');
title('Random data points clustered (own implementation)');
grid on;

% number of points in each cluster
subplot(2,2,2);
histogram(cluster);
axis tight;
[num,~] = histcounts(cluster);
yticks(round(linspace(0,max(num),k)));
xlabel('Clusters');
ylabel('Number of data points');
title('Histogram of the cluster points (own implementation)');

% visualize MATLAB's clustering
subplot(2,2,3);
scatter(xP,yP,200,cluster_m,'.');
hold on;
scatter(centr_m(:,1),centr_m(:,2),'xk','LineWidth',1.5);
axis([0 xMax 0 yMax]);
daspect([1 1 1]);
xlabel('x');
ylabel('y');
title('Random data points clustered (MATLABs implementation)');
grid on;

% number of points in each MATLAB cluster
subplot(2,2,4);
histogram(cluster_m);
axis tight;
[num_m,~] = histcounts(cluster_m);
yticks(round(linspace(0,max(num_m),k)));
xlabel('Clusters');
ylabel('Number of data points');
title('Histogram of the cluster points (MATLABs implementation)');